The difference in gain scores produced by traditional pretests and those produced by retrospective pretests when compared to posttest scores on the Science Teaching Efficacy Belief Instrument for preservice teachers was investigated in this study. Results indicated that gain scores using the traditional pretest produced significant improvement in one factor, but not in the other, which is congruent with most other results for this instrument found in the literature. However, gain scores produced by using retrospective pretests produced a greater magnitude of significance for the first factor, and also produced significant gains for the second factor. Follow-up interviews with participants provided evidence for greater internal validity for the retrospective pretest. Findings support the notion that retrospective pretests may produce gain scores with greater validity and greater statistical power.